Skip to content

Latest commit

 

History

History
102 lines (69 loc) · 4.54 KB

README.md

File metadata and controls

102 lines (69 loc) · 4.54 KB
Raw

Integration testing for Splunk Operator

Overview

This README describes the test framework, how to run the various tests and how to write additional tests.

Test framework

The test framework is based on Ginkgo/Gomega test framework. It is located in the test folder in the splunk-operator repository. The test framework has 2 main objects: TestEnv and Deployment.

TestEnv

The TestEnv object represents the test environment that tests are run against. The test environment is a Kubernetes namespace. When a TestEnv is created, it will create a namespace in the current Kubernetes cluster. The following resources are then created in that new namespace:

  1. Service Account
  2. Role
  3. RoleBinding
  4. Operator/Deployment

The tests are then run within the scope/context of the TestEnv object. Resources used by the tests are scoped to the TestEnv object (ie Kubernetes namespace). When the tests are completed, the TestEnv object is destroyed and the resources within are deleted.

Deployment

The Deployment object is simply to encapsulate what we are deploying within the TestEnv object and help with cleanup. We can deploy standalone, indexers, search heads, ...etc within a single Deployment object. A TestEnv can have 1 or more Deployments.

Running the tests

To run the tests, you will need to create or use an existing Kubernetes cluster. If you have a powerful machine and Docker engine installed, you can create a local Kubernetes cluster on top of Docker (Kubernetes-IN-Docker aka KIND cluster) for testing purpose.

To create a KIND cluster, run

make cluster-up (make cluster-down to tear down the cluster)

If you are using an existing cluster, you will need to install the enterprise CRDs

kubectl apply -f deploy/crds

To run the test,

make int-test

This calls the test/run-test.sh script. This test script will recursively runs all the tests in the test folder againsts the current Kubernetes cluster.

The test will output the results into a junit xml output file.

Note: Both make cluster-up and make int-test invokes the test/deploy-cluster.sh and test/run-test.sh scripts respectively. Both these scripts take in the env.sh script that defines various environment variables used to build the cluster and run the test.

  • SPLUNK_OPERATOR_IMAGE (Default: splunk/splunk-operator:latest): The Splunk Operator image to use for testing
  • SPLUNK_ENTERPRISE_IMAGE (Default: splunk/splunk:latest) : The splunk enterprise image to use for testing
  • CLUSTER_PROVIDER (Default: kind) : Supports creating eks or kind cluster
  • PRIVATE_REGISTRY (Default: localhost:5000) : Private registry to use push and pull the images used during testing

Note: To run a specific test, you can

  1. cd ./test/{specific-test} folder
  2. ginkgo -v -progress --operator-image=localhost:5000/splunk/splunk-operator:latest --splunk-image=localhost:5000/splunk/splunk:latest

Github actions

Smoke and integration tests will run on Github actions. This tests can be triggered on schedule or after a certain event occur depending of the content of the workflow files. To run the tests on different clusters platforms, you will need to define the following project environment variables. For AWS: AWS_ACCESS_KEY_ID AWS_SECRET_ACCESS_KEY CLUSTER_PROVIDER=[eks] ECR_REGISTRY

For Azure: STORAGE_ACCOUNT STORAGE_ACCOUNT_KEY CLUSTER_PROVIDER=[azure]

Writing tests

Ginkgo test framework has test suite and test spec ("It") files. A test suite file is the main test entry point and represents the test package. Each test folder is a test package and should have 1 test suite file and 1 or more test spec files. The test suite file to use to create and delete the TestEnv object. The test spec files are used to create and delete the Deployment object.

Adding a new test spec

  1. To add a new test spec, you can either add a new "It" spec in an existing test spec file or add a new test spec file. If you are adding a new test spec file, it is best to copy an existing test spec file in that suite and modified it.

Add a new test suite

  1. If you are adding a new test suite (ie you want to run the test in a separate TestEnv or k8s namespace), it is best to copy the test/example folder entirely and modify it

Notes

  1. As mentioned above, by default, we create a TestEnv (aka k8s namespace) for every test suite AND a Deployment for every test spec. It may be time consuming to tear down the deployment for each test spec especially if we are deploying a large cluster to test. It is possible then to actually create the TestEnv and Deployment at the test suite level.